AD is a devastating disorder of the aged population. Current treatments offer at best modest efficacy; thus, there is a need for a therapy able to alter the disease's pathophysiology (Hardy and Selkoe, 2002). The neuropathology consists of insoluble deposits of Aβ plaques, reactive gliosis, and intracellular neurofibrillary tangles of hyperphosphorylated tau (for reviews, see Price and Sisodia, 1998; Selkoe, 2000). However, the precise relationship between these events remains to be established (Mudher and Lovestone, 2002).

Aβ peptides containing 40 or 42 amino acids are produced by sequential cleavage of APP by β-secretase and then C-terminally by γ-secretase (Selkoe, 1999). The Aβ(42) peptide is highly amyloidogenic, driving amyloid fibril formation, which is considered the primary agent for direct amyloid toxicity and plaque formation (Maggio et al., 1992; Jarrett et al., 1993). Mutations in human APP around these secretase cleavage sites are associated with early onset familial AD, characterized by elevated levels of Aβ(42) (Price and Sisodia, 1998). Transgenic mice with mutant human APP demonstrate elevated production of Aβ(42), recapitulating the cerebral amyloidosis seen in AD (Games et al., 1995; Hsiao et al., 1996).

The amyloid deposition in these mouse brains is age-dependent, occurring slightly later than reported cognitive deficits (Janus and Westaway, 2001; Kawarabayashi et al., 2001; Lewis et al., 2004). Recently, a triple transgenic mouse revealed that the intracellular accumulation of the Aβ(42) peptide precedes extracellular plaque formation, correlating significantly with both the cognitive and electrophysiological dysfunction, which further implicates Aβ(42) as a key pathological event in this disease (Oddo et al., 2003). Consequently, much effort has focused on the inhibition of Aβ production, with one strategy being to inhibit γ-secretase, an enzyme complex composed of presenilin, nicastrin, APH-1, and PEN-2 (for review, see Haass, 2004). The development of selective γ-secretase inhibitors has allowed pharmacological reduction in Aβ production (for review, see Harrison et al., 2004).

γ-Secretase is a member of the I-CLiP protease family and cleaves a number of additional intramembrane substrates, including CD-44, Erb4, E-cadherin, Notch, and the Notch ligands Delta and Jagged 2 (for review, see Wolfe and Kopan, 2004). One of the best characterized is the transmembrane protein Notch, which is cleaved to become the transcriptionally active Notch intracellular domain (Hartmann et al., 2001; Wolfe and Kopan, 2004). Inhibition of its production has been identified as a potential concern for γ-secretase inhibitor therapy, because high levels of γ-secretase inhibition affect B- and T-cell maturation and ileal goblet cell formation (Searfoss et al., 2003; Wong et al., 2004).

Numerous studies have also demonstrated that Aβ is transported in and out of the brain by different mechanisms (for review, see Zlokovic, 2004). Determining in vivo changes of Aβ and amyloid deposition in the central nervous system resulting from chronic γ-secretase inhibition will further aid our understanding of these dynamics.

Previously, we characterized the effects of the novel potent, bioavailable γ-secretase inhibitor MRK-560 (Churcher et al., 2006) on diethylamine (DEA)-soluble Aβ in the brain and cerebrospinal fluid of the rat, demonstrating a dose-dependent effect on inhibition of Aβ production with a half-life suitable for once-a-day dosing. In addition, preliminary visual inspection of the gastrointestinal system in a 2-week dosing study of MRK-560 in the rat revealed no overt signs of toxicity (Best et al., 2006).

Thus, this study set out to determine the effect of chronic dosing of MRK-560 on Aβ levels and amyloid plaque formation in the Tg2576 mouse model of amyloid deposition. In addition, we examined peripheral tissues histologically for signs of toxicity that may be related to altered Notch signaling (Doerfler et al., 2001; Hadland et al., 2001; Searfoss et al., 2003; Wong et al., 2004).

Materials and Methods

Animals and Dosing. All procedures were conducted in accordance with the Animals (Scientific Procedures) Act of 1986 and its associated guidelines. Tg2576 transgenic mice (male and female) overexpressing human APP harboring the Swedish mutation (K670N, M671L; Hsiao et al., 1996) were bred in-house. All animals were maintained on a 12:12-h light/dark cycle with unrestricted access to food and water until use. Four-month-old mice were dosed orally for 1 week once a day with 3 mg/kg MRK-560 as a suspension in 0.5% methylcellulose at 10 ml/kg. For the chronic study, 12-month-old mice were dosed orally for 3 months once a day with 3 mg/kg MRK-560 as a suspension in 0.5% methylcellulose at 10 ml/kg.

Tissue Sample Preparation for 1-Week Study. Tg2576 mice were euthanized by stunning followed by decapitation at predetermined time points. Brains were removed, immediately frozen on dry ice, and stored at –80°C until use. The frozen brains were homogenized in 10 volumes (wt/v) of 0.2% DEA containing 50 mM NaCl, pH 10, and protease inhibitors (Complete; Roche Diagnostics, Mannheim, Germany) (Savage et al., 1998) and then centrifuged at 355,000g at 4°C for 30 min (Optima MAX ultracentrifuge; Beckman Coulter Fullerton, CA). The resulting supernatant was retained as the soluble fraction and neutralized by addition of 10% 0.5 M Tris-HCl, pH 6.8. Samples were frozen at –80°C awaiting analysis by immunoassay [in young plaque-free mice, DEA extraction yields similar levels of Aβ to the guanidine hydrochloride (GnHCl) method; Best et al., 2005].

Necropsy and Tissue Collection after Chronic Dosing Study. At necropsy, the brain was removed and divided into four quadrants. The frontal two quadrants, extending from the optic chiasm rostrally and including the frontal cortex, striatum, and olfactory bulbs, were frozen on dry ice. The remaining two caudal quadrants containing the hippocampus, overlying cortex, cerebellum, and brainstem were separated along the midline. The left hemisphere was frozen in isopentane held at –40°C on dry ice and then stored at –80°C. The remaining right caudal quadrant was immersion fixed in 10% neutral buffered formalin and processed in paraffin wax for serial sectioning in the sagittal plane. All remaining organs were dissected and immersion fixed in 10% neutral buffered formalin; selected organs were further processed and sectioned for subsequent analysis.

Immunoassay Analysis. The protocol used for extraction of “total” amyloid from the brain sections is based on the GnHCl extraction method described previously (Johnson-Wood et al., 1997). The frontal brain quadrants were homogenized in 10 volumes of 5 M GnHCl, 50 mM HEPES, pH 7.3, 5 mM EDTA plus 1× EDTA-free protease inhibitor cocktail (Complete). After mixing at room temperature for 3 h, the homogenate was diluted 10-fold into ice-cold 25 mM HEPES, pH 7.3, 1 mM EDTA, 0.1% bovine serum albumin plus 1× protease inhibitor cocktail and centrifuged at 16,000g for 20 min at 4°C. Aliquots of supernatant were stored at –80°C to prevent degradation. The biotinylated antibody 4G8 (Kim et al., 1988) was used in combination with the monoclonal antibodies G2-10 or G2-11 for detection of Aβ(40) or Aβ(42), respectively (Ida et al., 1996). These species reflect subpopulations of peptides with heterogeneous N termini encompassing at least the 4G8 epitope at residues 17 to 24. Analysis of the samples was performed using the SECTOR Imager 6000 (Meso Scale Discovery, Gaithersburg, MD), as described previously (Best et al., 2005).

Immunohistochemistry. A uniform random sample of sections representing 10 levels from the dorsomedial extent of the right caudal quadrant was selected for immunohistochemical labeling. In brief, sections were dewaxed and rehydrated through graded alcohols to phosphate-buffered saline. Immunohistochemical staining was performed in accordance with the manufacturer's instructions by using the anti-mouse IgG Vectastain Elite ABC kit and Mouse on Mouse blocking reagents (Vector Laboratories, Peterborough, UK). Aβ(40) was labeled using a mouse monoclonal antibody G2-10 at a dilution of 1:300. Sections were counterstained in Gill's hematoxylin (A. Menarini Diagnostics, High Wycombe, UK), dehydrated, cleared, and mounted.

Ligand Autoradiography. A frozen quadrant of brain containing the hippocampus was serial sectioned at 20 μm in the sagittal plan; three sections per slide were taken onto sterile SuperFrost Plus and the next three sections discarded. A uniform random sample (sections taken at a uniform interval to give 12 levels per animal) was taken for radiolabeled amyloid binding using a modified version of the protocol used by Maggio et al. (1992). The slides generated from the sectioning procedure were stored at –80°C until required. Sections were brought to room temperature and preincubated in 50 mM Tris, pH 7.4, 10 mM MnCl₂, 0.004% bacitracin, 0.002% chymostatin, and 0.004% leupeptin for 10 min before addition of radioligand at 50 pM (25 μCi; GE Healthcare, Little Chalfont, Buckinghamshire, UK). After incubation with the radioligand for 2 h, the slides were washed with 50 mM Tris-HCl, pH 7.4 (four 2-min washes at 4°C) and distilled water. The slides were then allowed to dry at room temperature. The fully dried slides were apposed to film (Kodak Biomax MR; Kodak Ltd., Hemel Hempstead, UK) along with radio-iodinated standards for 7 days (GE Healthcare). All films contained representatives of each treatment group to ensure standardization across the study. The film images for each group were captured and analyzed for plaque numbers digitally (Microdensitometry-Based Computer-Assisted Imaging Device-Analytical Imaging Station 6.0; Imaging Research, St. Catharines, ON, Canada). The mean plaque or “deposit” volume was calculated by dividing the stereologically derived total amyloid volume by the total number of plaques.

Histopathology. Hematoxylin- and eosin-stained sections (6 μm) of brain at five representative levels, and sections of ileum, spleen, and thymus, were examined by a blinded skilled pathologist for the qualitative evaluation of any treatment related changes.

Western Blotting. Brains from control and compound-treated Tg2576 mice were weighed and lysed in 15 volumes of ice-cold lysis buffer [20 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM sodium orthovanadate, 1 μg/ml leupeptin, and 1 mM 4-(2-aminoethyl)-bezenesulfonylfluoride] using a Teflon-coated pestle and mortar. Insoluble material was removed via centrifugation at 14,000 rpm for 5 min. The resulting supernatant was assayed for total protein content using the BCA microprotein assay (Perbio Science UK Ltd., Cramlington, UK) and mixed 1:1 with 2× sample buffer, and 30 μg was loaded per well onto 26-well 10% BisTris XT gels (Bio-Rad, Hemel Hempstead, UK). Electrophoresed proteins were transferred to Protran nitrocellulose membranes (Whatman Schleicher and Schuell, Dassel/Relliehausen, Germany). Primary immunodetection was via anti-glial fibrillary acidic protein (GFAP) (1:1000; DAKO UK Ltd., Ely, UK); specific bands were visualized using infrared dye-conjugated secondary antibodies and quantified using the Odyssey infrared imager (LI-COR Biosciences UK Ltd., Cambridge, UK).

Results

One-Week Dosing Study of MRK-560 in Young Tg2576 Mice. Based on previous measurements of the pharmacokinetic properties of MRK-560 (Best et al., 2006) and efficacy achieved in APP-YAC mice (Churcher et al., 2006), 3 mg/kg was administered p.o. to Tg2576 mice for 1 week (steady-state level) to determine the reduction of DEA-soluble brain Aβ(40) levels to establish reduction of Aβ for the chronic studies (Fig. 1). As expected, the reduction of these soluble Aβ(40) levels was fairly constant during the time course, with the integrated reduction over 24 h being 79%. Because in young plaque-free mice DEA-extracted Aβ should equate to GnHCl-extracted Aβ (Best et al., 2005), this dose of 3 mg/kg was chosen for chronic (3-month) dosing.

Effect of Chronic Administration of MRK-560 on Aβ Levels in Aged Tg2576 Mice. Animals were dosed from 12 months of age for 3 months. Chronic administration of 3 mg/kg/day MRK-560 did not lead to either increased mortality or deviation from control values for weight during the study period (data not shown). Initial levels of GnHCl-extracted Aβ(40) and Aβ(42) at 12 months of age were 1.47 ± 0.25 and 0.54 ± 0.07 nmol/g, respectively.

After 3 months of dosing, the levels of Aβ(40) in this 15-month-old group of animals receiving vehicle had increased to 9.63 ± 0.95 nmol/mg (Fig. 2A), an increase of approximately 6- to 7-fold compared with the 12-month-old animals. The animals receiving 3 mg/kg MRK-560 had levels of Aβ(40) of 5.48 ± 0.10 nmol/mg. This was a significant reduction of 43% compared with the 15-month vehicle group (Student's t test; p < 0.01). The levels of Aβ(42) (Fig. 2B) in the vehicle group were 3.90 ± 0.423 nmol/mg, also an increase of approximately 7-fold compared with the 12-month-old animals. In the group receiving 3 mg/kg MRK-560, the levels of Aβ(42) were 2.22 ± 0.22 nmol/mg. This was also equivalent to a 43% reduction compared with the 15-month-old vehicle group (Student's t test; p < 0.01).

Effect of Chronic Administration of MRK-560 on Plaque Numbers in Aged Tg2576 Mice. Evaluation of autoradiographic sections from the frozen caudal brain quadrant labeled with ¹²⁵I-Aβ(1-40) revealed a substantial decrease in the amount of binding in the inhibitor-treated versus vehicle-treated animals (Fig. 3, A versus B). Quantification of this binding demonstrated a significant 49% decrease in the number of amyloid deposits to which the ¹²⁵I-Aβ(1-40) was bound (Fig. 3C; Student's t test; p < 0.01). There was no significant change in the average volume of deposits compared across groups (Fig. 3D). This was derived by dividing the total amyloid volume by the deposit number. To confirm the effect seen with the ¹²⁵I-Aβ labeling, paraformaldehyde-fixed paraffin-embedded sections were immunolabeled for Aβ(40) using the N-terminal G2-10 antibody. The immunoreactivity also revealed a similar decrease in the number of plaque deposits in the γ-secretase inhibitor-dosed group (Fig. 3, E versus F).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Steady-state levels of MRK-560 after once a day dosing for 7 days reduce Aβ(40) levels for 24 h in Tg2576 mice. Time course graph of the effect of 3 mg/kg MRK-560 on DEA-extracted Aβ(40) levels in the brain. Animals were culled immediately 4, 8, 16, or 24 h after the last dose. Data points represent average Aβ(40) levels normalized to the vehicle levels ± S.E.M. (n = 3–5/data point).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Chronic dosing of 3 mg/kg MRK-560 once a day for 3 months reduces total Aβ levels in the brains of Tg2576 mice. Graphs represent effects of vehicle (open bars) versus 3 mg/kg MRK-560 (closed bars) on GnHCl-extracted Aβ(40) (A) and Aβ(42) (B) levels. Columns represent average Aβ levels (nanomoles per gram ± S.E.M.; n = 22–23 per column). **, p < 0.01.

Effect of MRK-560 on Associated Gliosis in Aged Tg2576 Mice. Reactive astrocytosis is known to be associated with an increase in Aβ levels in AD. To determine whether this relationship was maintained in this study, Western blot analysis was performed for GFAP on the brains of the mice (Fig. 4). The Western blots demonstrated a decrease in the levels of GFAP in response to the inhibitor. When the blots were quantified, a significant decrease of 27% was observed (Student's t test; p < 0.05).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Plaque burden is reduced in the brains of Tg2576 mice treated with 3 mg/kg MRK-560 once a day for 3 months. Autoradiographs of posterior quadrants of the brain, including areas such as cortex, hippocampus, and cerebellum for the radioligand binding of ¹²⁵I-Aβ to plaques in mice treated with either vehicle (A) or MRK-560 (B). Note main localization of binding to cortex and hippocampal areas and marked reduction of binding in MRK-560 group. C, quantification of binding (expressed as number of plaques per cubic millimeter of cortex), demonstrating a significant reduction in the MRK-560-treated group (p < 0.01) (n = 12). D, quantification of the mean deposit volume showing that there was no overall difference in the size of the plaques between the vehicle and MRK-560 group (n = 12). Immunohistochemical labeling of plaques with G2-10 antibody in the cortex and hippocampus in the vehicle- (E) and MRK-560 (F)-dosed group revealing a reduction in the number of plaques in the MRK-560-treated group. Inset shows close-up of a labeled plaque.

Histopathology after Chronic Administration of MRK-560 in Peripheral Tissues of Tg2576 Mice. One of the main concerns regarding the use of γ-secretase inhibitors is the off-target effects on substrates and organs not directly related to the neuropathology of AD (Searfoss et al., 2003; Wong et al., 2004). Of these, the ileum, thymus, and spleen have been picked out as potential targets and markers of this side effect liability. To determine whether there were any off-target effects caused by this dosing regime, sections of ileum (Fig. 5, A and B), spleen (Fig. 5, C and D), and thymus (Fig. 5, E and F) were examined for any microscopic changes relating to the administration of MRK-560. There were no discernible differences between any of these tissue sections; in the case of the ileum, this was particularly related to the number of goblet cells that did not change along with the tissue architecture. The same picture was seen in the thymus and spleen. In the spleen, there was no discernible decrease in the size of the marginal zone cells (lighter staining; Fig. 5, C and D) between the vehicle and MRK-560 group. In the thymus, the expected atrophy of the cortical zone was not seen, with a clear demarcation between the darker stained cortical zone and lighter stained medullary zone seen in both vehicle and MRK-560 groups (Fig. 5, E and F). Microscopically, although the number of cells was not quantified, no morphological differences were seen between cells from the vehicle and the MRK-560 groups of animals in either the spleen or thymus. These findings indicated no deleterious effects of MRK-560 on the cell differentiation pathways in these peripheral tissues.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

GFAP staining is reduced in the brains of Tg2576 mice treated with 3 mg/kg MRK-560 once a day for 3 months. Top, representative Western blot of three animals from vehicle- and MRK-560-treated groups revealing decrease in GFAP staining (arrow) indicative of reactive gliosis. Graph represents quantification of blots demonstrating significant reduction of GFAP in mice treated with MRK-560 versus vehicle-treated animals (Student's t test; p < 0.05) (n = 14).

Discussion

The ability of γ-secretase inhibitors to lower levels of Aβ after acute in vitro and in vivo exposure has been reported in various model systems (for review, see Churcher and Beher, 2005). There are also reports investigating the effects of these types of compounds on Aβ in old transgenic mice following chronic administration (Barten et al., 2005). Our data represent the first demonstration that chronic administration of a γ-secretase inhibitor during a period of accelerated plaque deposition can significantly reduce the initiation and extent of these lesions without significant effects on the peripheral tissues.

Using a combined approach of biochemical and histological analysis, we have demonstrated a significant decrease in both the number and extent of amyloid deposits in the cortex of inhibitor-treated mice compared with their vehicle-treated controls. These differences were significant in all three assay systems. Immunohistochemical and ¹²⁵I-Aβ detection of plaques revealed a significant decrease in the overall fraction occupied by amyloid.

Interestingly, individual deposits in treated mice did not differ in average size from those seen in equivalent vehicle-dosed animals. This would imply that reduction of Aβ by 40 to 50% has a significant effect on the initiation of the formation of plaques rather than their growth or clearance. Hence, it is likely that once seeded, amyloid concentrations remain sufficient for plaques to increase in size at a rate similar to that seen under control conditions. This may be especially true in the presence of lower overall numbers of deposits acting as a sink for the soluble amyloid pool. Recent reports have demonstrated that intracellular accumulation of Aβ(1-42) can precede the appearance of extracellular plaques (Oddo et al., 2003) and that dense aggregates of Aβ(1-42), considered to be the focus for further deposition, are deposited before the appearance of diffuse amyloid. Given the clear reduction in plaque-deposited Aβ(40), it would also be interesting to use thioflavin staining and immunohistochemical staining for Aβ(42) to determine whether dense core plaques numbers are affected differently by MRK-560 treatment. Thus, taking these observations into account, sustained γ-secretase inhibition during a period when the accumulation and aggregation of intraneuronal amyloid is causing cellular and synaptic degeneration may be critical.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Chronic dosing of 3 mg/kg MRK-560 once a day for 3 months does not induce abnormal changes in peripheral tissues of Tg2576 mice. Representative tissue sections from ileum (A and B), spleen (C and D), and thymus (E and F), demonstrating that treatment with MRK-560 (B, D, and F) did not induce pathological changes in these tissues compared with the vehicle-treated group (A, C, and E), despite lowering Aβ-containing plaques in the brain.

Previously, we have established that GnHCl extraction (Johnson-Wood et al., 1997; Lewis et al., 2004) of brain homogenates extracted a significant proportion of the total amyloid pool in this mouse model. This extraction protocol solubilizes most of the amyloid except some plaque core and vascular amyloid (D. W. Smith, unpublished observation). At 15 months, the soluble pool (as reflected by sodium dodecyl sulfate or DEA extraction) makes up approximately 10% of the Aβ pool extracted by formic acid or GnHCl (Kawarabayashi et al., 2001). The lowering in plaque number as assessed by ¹²⁵I labeling agreed well with the decrease in GnHCl-extracted Aβ for both Aβ(40) and Aβ(42) in the sampled brain containing the olfactory bulb, frontal cortex, and striatum, suggesting that these biochemical measures are an accurate reflection of overall amyloid load, including plaque number.

Plaque formation and maturation are normally accompanied by inflammatory processes. Initial findings using Western blot analysis to assess GFAP immunoreactivity demonstrated a reduction in the Tg2576 mice chronically treated with the γ-secretase inhibitor. Because these reductions probably reflect the decrease in the number of lesions rather than a generalized decrease in the activation of glia, detailed immunohistochemistry with morphological evaluation and quantification would be required, because there is no evidence that an individual deposit, once formed, does not elicit the typical glial response and subsequent damage to the surrounding neuropil.

A key to establishing the clinical viability of γ-secretase inhibition is the demonstration of a safety window between the effects on processing of APP and the effect on alternative substrates for γ-secretase, particularly the Notch family of receptors. It has been suggested that this lack of substrate specificity has the potential to lead to adverse effects.

Much is known about the developmental importance of Notch in cell fate determination. Removal of the Notch pathway transcription factor CSL/RBP-J in transgenic mice generated defects in the gastrointestinal tract with a similar phenotype being achieved using the γ-secretase inhibitor dibenzazepine (van Es et al., 2005). This effect was also demonstrated in rats after 4 to 5 days of dosing with the γ-secretase inhibitors compound X and dibenzazepine (Searfoss et al., 2003; Milano et al., 2004). The effects manifested themselves grossly as an increase in the gastrointestinal weight associated with distension of the small and large intestines. Microscopically, goblet cell metaplasia was observed in the ileal lumen along with abnormal villus architecture. Another organ affected was the spleen, where a decrease in the number of marginal zone cells was observed. In TgCRND8 mice dosed with the γ-secretase inhibitor LY-411575 for 2 weeks, atrophy of the thymus was reported with doses that gave between 60 and 80% reduction of Aβ(40) and Aβ(42) (Wong et al., 2004).

Recently, a number of reports have revealed compounds that can distinguish between the proteolytic activities of APP and Notch. These compounds are not classical γ-secretase inhibitors [i.e., they do not inhibit the production of Aβ(1-40) and Aβ(1-42) equally]; rather they are modulators of γ-secretase cleavage specificity in that they favor the reduction of Aβ(42) production. Such compounds include a subset of NSAIDs, which preferentially reduced Aβ(42) production without seeming to affect Aβ(40) or Notch processing (Weggen et al., 2001, 2003). However, evidence is contradictory as to whether these compounds are effective in vivo (Eriksen et al., 2003; Lanz et al., 2005). Likewise, of the pan inhibitors, the aryl sulfonamide BMS-299897 seemed to have a 15-fold separation between APP and Notch processing in human embryonic kidney 293 cells (Barten et al., 2005). In vivo, separate studies in rats and Tg2576 mice revealed there were no adverse phenotypes in gut, spleen, or thymus when BMS-299897 was dosed for a period between 5 days and 2 weeks (Milano et al., 2004; Barten et al., 2005). However, brain Aβ reduction was not investigated in the rats and BMS-299897 was unable to lower brain Aβ(40) in the aged plaque-bearing Tg2576 mice (Barten et al., 2005). Thus, in these studies no window between amyloid production and Notch cleavage has been unequivocally demonstrated in vivo.

We conducted detailed histopathological evaluation of a number of potential target organs in the mice that had been treated for 3 months. Analysis of the gastrointestinal tract (represented by the ileum) revealed no gross changes in weight or appearance. Microscopically, there were no differences in goblet cell numbers or general villus architecture between the vehicle-dosed animals and the animals receiving MRK-560. The spleen presented no change in the marginal cell zone between the groups; this is consistent with the previous data for BMS-299897. Finally, evaluation of the thymus demonstrated no atrophy or abnormal changes in contrast to the changes caused by LY-411575 in the TgCRND8 mice (Wong et al., 2004).

When taken together, our data demonstrate that chronic γ-secretase inhibition can significantly reduce the induction of amyloid deposits together with the associated inflammatory changes in the brains of Tg2576 mice. This can be achieved at levels of inhibition that do not seem to induce histopathological changes in the brain or peripheral organs that would be associated with inhibition of processing of alternative substrates by this enzyme complex.

In vitro, there seems to be little or no separation for MRK-560 between the inhibition of the APP and Notch processing pathway (Churcher et al., 2006). Using human embryonic kidney 293 cells stably coexpressing APP and NotchΔE (Lewis et al., 2003), MRK-560 had IC₅₀ values for APP and Notch cleavage of 4.32 and 3.44 nM, respectively. In this same assay, LY-411575 gave values of 0.119 nM for APP and 0.129 nM for NotchΔE (Lewis et al., 2003). Because LY-411575 has been shown to cause severe peripheral organ toxicity in TgCRND8 mice (Wong et al., 2004), a full characterization of the therapeutic window of MRK-560 in vivo in this dosing paradigm is needed to be able to compare these inhibitors.

The mechanism whereby MRK-560 has beneficial effects on amyloid plaque deposition in the absence of toxicity related to changes in the Notch signaling pathway remains to be elucidated. However, the composition of the γ-secretase complex can vary from tissue to tissue with respect to differences in the APH-1 isoforms (Serneels et al., 2005). This offers the intriguing possibility that certain structural classes of γ-secretase inhibitors may preferentially affect brain but not peripheral γ-secretase complexes. This idea could be further investigated using other endpoints such as human endostatin expression (Milano et al., 2004). This would then refocus attention on γ-secretase inhibitors as a possible therapy for AD, an approach that until now seemed to be limited due to mechanism-based peripheral toxicity (Searfoss et al., 2003; Wong et al., 2004).